In Situ Laser Raman Spectroscopy Studies of VPO Catalyst

Structural Characteristics, Stability, and Activity of (VO)2P2O7 and VO(PO3)2 Catalysts in p-Cymene Liquid-Phase Oxidation. Peter R. Makgwane , Ernst ...
81 downloads 0 Views 174KB Size
Download PDF
J. Phys. Chem. B 1999, 103, 9459-9467

9459

In Situ Laser Raman Spectroscopy Studies of VPO Catalyst Transformations Zhi-Yang Xue and Glenn L. Schrader* Department of Chemical Engineering and Ames Laboratory-USDOE, Iowa State UniVersity, Ames, Iowa 50011 ReceiVed: April 19, 1999; In Final Form: July 14, 1999

VPO catalyst transformations were investigated using in situ laser Raman spectroscopy. During reductionoxidation step changes, (VO)2P2O7 was readily converted to RII-, δ-VOPO4, and ultimately to β-VOPO4 in O2/N2; these V5+ phases were eliminated in n-butane/N2. A wet N2 feed (5-10% H2O in N2) transformed (VO)2P2O7 and RI-, RII-, β-, δ-, γ-VOPO4 to V2O5 at temperatures above 400 °C. The presence of water vapor facilitated the loss of oxygen atoms involved in V-O-P bonding, and separated vanadium oxide and phosphorus oxide species were formed. The isolated vanadium oxide species could be transformed to V2O5; phosphorus species likely diffused from the catalyst lattice in the form of acid phosphates.

1. Introduction Vanadium phosphorus oxide (VPO) catalytic materials and their precursors involve many known crystalline phases. Compounds with a vanadium valence of +4 include vanadium hydrogenphosphate hydrates (VOHPO4‚xH2O, x ) 0.5, 1, 2, 3, 4) and vanadyl pyrophosphate [(VO)2P2O7]. Vanadyl orthophosphate phases (RI-, RII-, β-, δ-, γ-VOPO4, and VOPO4‚2H2O) have a vanadium valence of +5. Phase transformations between these VPO materials are rather complicated, as reported in the literature (Figure 1). In general, phase transformations among compounds of the same vanadium valence can be induced by thermal treatment, whereas conversions between (VO)2P2O7 and VOPO4 phases require a reducing or oxidizing atmosphere. In addition, the effect of water on catalyst performance has been investigated.1,2,3 Water is an inescapable product in n-butane selective oxidation to maleic anhydride. In fixed bed processes, with low n-butane feed concentrations (e.g., 1.5% n-butane in air), the reactor outlet water vapor fraction can approach 3%.1 Researchers at DuPont4 and Exxon5,6 were among the first to synthesize the hemihydrate phase VOHPO4‚0.5H2O and to determine its crystal structure. Vanadyl hydrogenphosphate tetrahydrate (VOHPO4‚4H2O) was obtained by exposing VOHPO4‚0.5H2O to room conditions for 4 months.6 Later Amoro´s and co-workers7 successfully synthesized a series of vanadyl hydrogenphosphate hydrates (R-, β-VOHPO4‚2H2O, VOHPO4‚3H2O, VOHPO4‚4H2O) using solutions with specific acetone/water ratios. These materials can ultimately be converted via two pathways (Figure 1) to (VO)2P2O7, which has been recognized as the important catalytic phase for commercial n-butane oxidation under steady-state conditions. VOHPO4‚ 0.5H2O is the direct precursor to (VO)2P2O7 for both pathways as revealed by XRD, neutron thermodiffractometry, and TGADSC.7-9 When VOHPO4‚0.5H2O is treated in an oxidative atmosphere, V5+ phases can be formed. Bordes et al.10 have prepared δ-VOPO4 (450 °C in air) and γ-VOPO4 (680 °C in oxygen) by thermal dehydration of VOHPO4‚0.5H2O. Hutchings et al.11 monitored the transformation of VOHPO4‚0.5H2O in 1.5% n-butane/air in situ using Raman spectroscopy. Following a structural disordering of VOHPO4‚0.5H2O at around 370 °C (revealed by an absence of Raman bands), (VO)2P2O7 and RII-, δ-, γ-VOPO4 were produced.

Figure 1. Diagram of phase transformations among VPO catalysts and precursors reported in the literature.

(VO)2P2O7 has been reported to transform to the β-VOPO4 phase under reactive conditions at about 760 °C in oxygen.12 Gai et al.13 probed the behavior of vanadyl pyrophosphate during a 312 h treatment in steam (31% or 42% water vapor fraction) at near 390 °C using in situ environmental high resolution electron microscopy (EHREM). An increase in the number of defects caused disorder in the catalyst. The resulting material after a prolonged calcination in steam was a “new, slightly anion-deficient phase”. The kinetic study using the steam-treated catalyst showed significantly lower reactivity and selectivity for maleic anhydride. β-VOPO4 is considered to be the most stable phase among anhydrous orthophosphates. Bordes et al.10,14 have reported that high temperature thermal treatment (e.g., 750 °C in N2) initiated the following transformations: RI- f RII- f β-VOPO4, or δf γ- f β-VOPO4. The conversion of β-VOPO4 to (VO)2P2O7 has been reported to occur at 400 °C using in situ laser Raman spectroscopy in 1% 1-butene/air flow.12 Schrader et al.15,16 also observed the reversible conversion of β-VOPO4 to (VO)2P2O7 at 500 °C in 2% n-butane/air. In situ Raman spectroscopic studies have indicated that δ-VOPO4 can transform to RIIVOPO4 in 2.4% n-butane/air flow at near 400 °C.17 RII-VOPO4, on the other hand, was transformed to γ-VOPO4 when heated in N2 flow at 700 °C.18

10.1021/jp9912713 CCC: $18.00 © 1999 American Chemical Society Published on Web 10/12/1999

9460 J. Phys. Chem. B, Vol. 103, No. 44, 1999 Using Raman spectroscopy, Ben Abdelouahabe et al.19 showed that hydration of RI-, RII-, δ-, and γ-VOPO4 formed VOPO4‚2H2O at room temperature, demonstrating that water molecules can diffuse into the catalyst lattice and alter the catalyst structure. At higher temperatures (>80 °C), the resulting VOPO4‚2H2O was converted to RI-VOPO4. DuPont’s process for n-butane selective oxidation using a circulating fluidized bed (CFB) reactor20 has raised new questions regarding the effect of transient operation conditions on the catalyst activity and VPO catalyst transformations. We have reported preliminary studies of VPO catalyst phase transformations under transient operation conditions:21 an industrial (VO)2P2O7 catalyst was exposed to a cycle of oxidizing (10% O2/N2) and reducing (2% n-butane/N2) conditions. In situ Raman data showed evidence for the formation of several vanadium orthophosphate phases. The current study has continued these efforts to characterize VPO catalyst phase transformations under transient conditions. Our new studies have again focused on (VO)2P2O7 since this phase is regarded as being crucial for industrial processes to convert n-butane to maleic anhydride.22 In addition, this phase appears to have a central role for several pathways involved in VPO phase transformations (Figure 1). Most previous investigations of the effect of water on catalyst performance suggested increased selectivity for maleic anhydride and decreased n-butane conversion.1,2,3 These effects have been attributed to (1) competitive adsorption of oxygen and water molecules on the catalyst surface; (2) preferential alteration in reaction rates of certain elementary steps induced by the participation of water in the reaction; or (3) changes in catalyst surface area. The possibility of water vapor inducing phase transformations in VPO materials has not been clearly determined. Studies conducted by Gai et al.13 have indicated that the presence of water vapor was detrimental to both selectivity and reactivity; structural changes were observed to occur for (VO)2P2O7. 2. Experimental Section 2.1. Catalyst Preparation. VPO catalytic materials were provided by DuPont. VOHPO4‚0.5H2O was synthesized in an organic media:23 V2O5 was reduced to form a VO2+ species by refluxing in a mixture of isobutanol and benzyl alcohol, and the desired amount of phosphorus was then added as anhydrous H3PO4. This precursor was calcined in air and activated in 1.5% n-butane-10% O2-balance N2 to produce (VO)2P2O7. β-VOPO4 was synthesized by reacting VOF3 with tristrimethylsilyl phosphate in acetonitrile. δ- and γ-VOPO4 were obtained by the oxydehydration of VOHPO4‚0.5H2O in dry oxygen or N2. Thermal treatment of VOPO4‚2H2O produced RI-VOPO4. V2O5 powder used as a reference was purchased from Fisher Scientific (purity 99.9%). 2.2. Catalyst Characterization. X-ray powder diffraction patterns of the catalysts were recorded with a Scintag XDS 2000 diffractometer using CuKR radiation. Infrared spectra of samples pressed as KBr pellets were obtained using a Nicolet model 60SX FTIR spectrometer. The Spex Triplemate laser Raman spectrometer was coupled to a cryogenic charge coupled detector (Princeton Instruments model LN/CCD). The laser source was a Spectra-Physics model 164 Argon ion laser operating at 514.5 nm. Spectra were obtained using a laser power of 40 mw at the source. The intensity of the incident laser beam on the sample was approximately 40% of the source power. Data acquisition was for 3 min or less since previous experiments confirmed that the VPO catalysts were stable under these conditions.21

Xue and Schrader

Figure 2. (VO)2P2O7 phase transformations in the redox (30 min air30 min n-butane/N2) step change reaction cycles at 400 °C: (a) spectrum taken after (VO)2P2O7 being treated in n-butane/air for 3 h; (b) and (c) after the oxidation step and the reduction step of the second redox cycle, respectively; (d) and (e) after the oxidation step and the reduction step of the third redox cycle, respectively; (f) and (g) after 15 and 30 min in the oxidation step of the fourth redox cycle; (h) and (i) after 15 and 30 min in the reduction step of the fourth redox cycle; (j) and (k) after 15 and 30 min in the oxidation step of the fifth redox cycle.

For in situ laser Raman characterization, a cell was constructed from MACOR machineable ceramic (a detailed description of the reactor cell has been given elsewhere21). A gas delivery system could direct a continuous flow of n-butane/air, n-butane/N2, air, or N2 into the cell; or, step changes between n-butane/N2 and air could be introduced. Research purity n-butane (Matheson), ultrapure N2 or air (Air Products) were used in the experiments. A saturator could be connected into the system for addition of water vapor. The concentration of water vapor (5-10 vol %) was controlled by adjusting the temperature of the saturator. Tylan mass flow controllers regulated the gas flow rates. The n-butane concentration was 2 vol % for continuous flow and 5 vol % for step change studies. The total volume flow rate was 50 sccm for all studies. 3. Results 3.1. Catalyst Characterization. X-ray diffraction data and the laser Raman spectra of all materials [V2O5, (VO)2P2O7 and RI-, RII-, β-, δ-, γ-VOPO4] were in good agreement with the published data.17,24 Only one material, VOPO4‚2H2O, was observed to be very sensitive to the laser exposures used in this study and was readily dehydrated to form RI-VOPO4. 3.2. (VO)2P2O7 Phase Transformations in OxidationReduction Cycles. (VO)2P2O7 was treated in a series of 30 min air-30 min 5% n-butane/N2 cycles at 400 °C. Figure 2a is the Raman spectrum of (VO)2P2O7 after being treated in n-butane/ air for 3 h at 400 °C. The first oxidation-reduction cycle did not induce phase changes (spectra not shown). In the second

VPO Catalyst Transformations

Figure 3. (VO)2P2O7 phase transformation in the wet feed [2% n-butane/(5%-10%) water vapor/air] at 450 °C: (a) spectrum taken at RT; (b) after 2 h in the wet feed; (c) after 5 h in the wet feed; (d) after the catalyst was subsequently cooled to RT; (e) taken at RT after 70 h thermal treatment at 450 °C.

cycle, two bands due to RII-VOPO4 (1090 and 995 cm-1) emerged during the oxidation step (Figure 2b). RII-VOPO4 also has a band at 945 cm-1, which was likely obscured by the intense (VO)2P2O7 band at 932 cm-1. In the following reduction step, the RII-VOPO4 bands disappeared and the spectral features were typical for (VO)2P2O7 (Figure 2c). Some changes, however, were apparent in the relative intensities of the bands. After the oxidation step of the third cycle, additional bands appeared at 1021 and 595 cm-1 and in the region of 1070-1098 cm-1 (Figure 2d). These bands (characteristic of δ-VOPO4) were eliminated in the subsequent reduction (Figure 2e). Figure 2f, g were collected at 15 and 30 min following an oxidation step. Bands near 1090, 1020, 995, 650, and 594 cm-1 emerged and intensified. In the subsequent reduction, the intensity of these bands diminished (Figure 2h, i). The catalyst was again oxidized for 30 min when Figure 2j, k were collected at 15 and 30 min, respectively, into the oxidation step. Bands other than those due to (VO)2P2O7 included a broad band in the region from 1060 to 1100 cm-1 with two maxima at 1090 and 1070 cm-1; a band centered at 995 cm-1; a shoulder band around 890 cm-1; bands at 645 and 590 cm-1; and a broad band in the region of 430360 cm-1. Bands at 1090 and 995 cm-1 were due to RII-VOPO4; those at 1070, 890, 649, 590, and 425 cm-1 clearly indicated the existence of β-VOPO4. 3.3. (VO)2P2O7 Transformations in Wet Feeds. In Figures 3-11, the results from a series of experiments involving the effect of water vapor on catalyst stability are shown. Figure 3 provides data for the transformation of (VO)2P2O7 in 2% n-butane/5%-10% water vapor/air at 450 °C. Figure 3a is a spectrum of (VO)2P2O7 at room temperature. The temperature of the catalyst was raised by 5 °C/min to 450 °C. After 2 h,

J. Phys. Chem. B, Vol. 103, No. 44, 1999 9461

Figure 4. (VO)2P2O7 phase transformation in the wet-dry feed [2% n-butane/(2%-10%) water vapor/air-2% n-butane/air] step change reaction cycles at 450 °C: (a) spectrum taken after (VO)2P2O7 being treated in wet feed for 4 h; (b) 2 h in the following dry feed treatment; (c) after 5 h in dry feed; (d) 1 h in the following wet feed treatment; (e) after 6 h in wet feed.

distinctive bands emerged at 994 cm-1 (with an apparent shoulder band around 1020 cm-1) and 141 cm-1; less intense bands were present near 700, 285, and 407 cm-1 (Figure 3b). These bands indicated the formation of a new phase which became more apparent later during the experiment (Figure 3c). When this material was cooled to room temperature, the bands due to the new phase disappeared and those due to (VO)2P2O7 were recovered, although the relative intensities of the bands were significantly altered (Figure 3d). However, if the catalyst was treated in the gas flow described above for an extended period of time (several days), the transformation was stabilized even after cooling to room temperature (Figure 3e). The new Raman bands are characteristic of V2O5. (VO)2P2O7 was converted to V2O5 similarly in wet N2 (5-10% water vapor/ N2). Step change studies alternating between wet feed (2% n-butane/5-10% water vapor/air) and dry feed (2% n-butane/ air) were also conducted to further investigate the effect of water vapor. After (VO)2P2O7 had been treated in wet feed for several hours at 450 °C (Figure 4a), a mixture of V2O5 (bands at 992, 289, 144 cm-1), (VO)2P2O7 (band at 933 cm-1), and β-VOPO4 (bands at 1063, 891, 645, and 590 cm-1) was present. When the feed was then changed to dry feed (Figure 4b), the intensity of signals from V2O5 and β-VOPO4 decreased relative to the intensity of the (VO)2P2O7 band at 933 cm-1 after 2 h. After 5 h in dry feed, only bands due to (VO)2P2O7 were observed (Figure 4c). Wet feed was then reintroduced, and bands from (VO)2P2O7 disappeared after 1 h. A broad band with two maxima at 1020 and 992 cm-1 and a band near 144 cm-1

9462 J. Phys. Chem. B, Vol. 103, No. 44, 1999

Xue and Schrader

Figure 6. RII-VOPO4 phase transformations in wet N2 (5-10% water vapor/N2) at 450 °C: (a) spectrum taken at RT; (b) after 1 h at 450 °C; (c) after 3 h at 450 °C; (d) after the temperature being lowered to RT. Figure 5. γ-VOPO4 phase transformations in wet N2 (5-10% water vapor/N2) at 450 °C; (a) spectrum taken at RT; (b) after being heated to 450 °C; (c) after 2 h at 450 °C; (d) and (e) spectra taken after the temperature being lowered to 200 °C and 70 °C, respectively.

became the major features (Figure 4d); further reaction in the wet feed confirmed the formation of V2O5 (Figure 4e). 3.4. VOPO4 Transformations Under the Effect of Water Vapor. Various VOPO4 phases were tested in wet N2 (5-10% water vapor/N2) at temperatures of 400-600 °C. Figure 5 shows the conversion of γ-VOPO4 phase to V2O5 at 450 °C. Figure 5a is the spectrum of γ-VOPO4 at room temperature. This phase apparently experienced structural disordering after being heated to 450 °C (Figure 5b). The spectrum showed only a weak band centered at 1020 cm-1. Within 2 h, strong peaks at 993, 408, 285, and 144 cm-1 and less intense bands at around 700, 480, and 400 cm-1 indicated the transformation of γ-VOPO4 to V2O5 (Figure 5c). Figure 5d and e were collected after the temperature was subsequently lowered to 200 °C and 70 °C. Intensification of the bands due to γ-VOPO4 was observed. Apparently, lowering the temperature reversed the catalyst transformation, and γ-VOPO4 was partially recovered. When RII-VOPO4 was treated in wet N2 feed at 450 °C, similar behavior was observed. Figure 6a is the spectrum of RII-VOPO4 at room temperature. After 1 h in wet N2 at 450 °C, all bands due to RII-VOPO4 diminished in intensity. Bands at 993 cm-1 (with shoulder at 1016 cm-1), 407, 285, and 143 cm-1 indicated the generation of V2O5 (Figure 6b, c). Similar to the observation in the previous case (1) a band at near 1016 cm-1 was observed early in the transformation (Figure 6b) and (2) crystalline RII-VOPO4 was partially recovered when the temperature was lowered to room temperature (Figure 6d). Similar studies for β-VOPO4 at 450 °C did not result in detectable transformations. At 600 °C, however, changes were readily observed (Figure 7). In the early stage of exposure to wet N2, only a broad band was observed in the range of 1020-

Figure 7. β-VOPO4 phase transformations in wet N2 (5-10% water vapor/N2) at 600 °C: (a) spectrum taken at RT; (b) after being heated to 600 °C; (c) after 2 h at 600 °C; (d) spectrum taken at RT after a thermal treatment for 48 h.

990 cm-1 with two maxima at 1016 and 993 cm-1 (see Figure 7c). Bands due to V2O5 evolved after several hours but this change could be reversed if the temperature was lowered. The catalyst was also treated in wet N2 at 600 °C for an extended time period. The resulting material was shown to be V2O5 at room temperature (Figure 7d).

VPO Catalyst Transformations

Figure 8. δ-VOPO4 phase transformations in wet N2 (5-10% water vapor/N2) at 450 °C: (a) spectrum taken at RT; (b) after being heated to 450 °C; (c) after 1 h at 450 °C; (d) spectrum taken at RT after a thermal treatment for 48 h.

Figure 8 illustrates the transformation of δ-VOPO4 to V2O5 at 450 °C in wet N2 feed. Comparison of the spectrum collected at room temperature (Figure 8a) with that collected at 450 °C (Figure 8b) reveals that the latter possessed both bands for δ-VOPO4 and bands at 996 and 146 cm-1 which indicated the existence of V2O5. Raman signals from δ-VOPO4 disappeared later in the process (Figure 8c). After an extended thermal treatment in wet air, the resulting V2O5 contained a small amount of β-VOPO4, indicated by bands at 1071, 895, and 652 cm-1. RI-VOPO4 has also been converted to V2O5 (Figure 9) in wet N2 feed. Figure 9a shows a well-crystallized RI-VOPO4 which later was gradually converted to V2O5 at 450 °C (Figure 9b, 9c). The spectrum of the postreaction catalyst (Figure 9d) exhibited bands only due to RI-VOPO4. In most cases, VOPO4 phases were converted to β-VOPO4 under thermal treatment in wet air. Figure 10 demonstrates the direct conversion from δ-VOPO4 to β-VOPO4. The starting material (Figure 10a) was largely δ-VOPO4 (bands at 1198, 1088, 1019, 942, 588 cm-1) with trace amounts of RII-VOPO4 (bands at 992 cm-1 and lower wavenumber bands such as 469, 152 cm-1, etc.). At 450 °C in wet air, bands due to the original material decreased in intensity while bands due to β-VOPO4 (1068, 987, 894, 651, and 433 cm-1, etc.) emerged and intensified (Figure 10b-e). The final material was predominantly β-VOPO4. Direct conversion of RI-VOPO4 to β-VOPO4 in wet air to 450 °C is shown in Figure 11. The decrease in the intensity of bands due to RI-VOPO4 and growth of those due to β-VOPO4 occurred during the process. Similar observations have been experienced for RII- and γ-VOPO4 phases. In contrast to the use of wet N2, no intermediate phase was detected for wet air studies. In addition, no spectra had a complete absence of Raman bands.

J. Phys. Chem. B, Vol. 103, No. 44, 1999 9463

Figure 9. RI-VOPO4 phase transformations in wet N2 (5-10% water vapor/N2) at 450 °C: (a) spectrum taken at RT; (b) after being heated to 450 °C; (c) after 2 h at 450 °C; (d) spectrum taken after the temperature being lowered to RT.

4. Discussion In situ LRS studies of the thermal treatment of various VPO phases in wet N2 reveal common behavior. Raman bands due to the VPO phases decreased in intensity at elevated temperatures (e.g., >400 °C), which eventually led to a complete loss of the Raman signal prior to the emergence of a weak band near 1016-1020 cm-1. The 1016-1020 cm-1 band was accompanied by the gradual appearance of a band at 993 cm-1. Other bands at 700, 526, 480, 407, 302, 283, 196, and 147 cm-1 intensified together with the 993 cm-1 band and became distinct with further wet N2 treatment. These bands can be unequivocally assigned to V2O5.25 The conversion of the VPO catalysts to V2O5 was reversible at lower temperatures: a partial recovery of the initial VPO catalyst was apparently possible. Nevertheless, an extended thermal treatment in wet N2 eventually converted VPO phases irreversibly to V2O5. The weakening and eventual loss of Raman signals early in the conversion is likely due to structural disorder. It is known for a periodic structure, the light scattering process satisfies the law of momentum conservation

k ) k0 + q where k0, k, and q are the wave vectors for the incident photon, the scattered photon, and the phonon being adsorbed or emitted.26 Structural disorder breaks the translational symmetry in crystalline materials and the momentum conservation law does not hold. As a result, phonons with various frequencies are involved in the scattering process which increases the background in Raman spectra and broadens the Raman signals. Scheetz and White27 suggested two types of disorder: those induced by electronic irregularity due to misorientation of nonspherical structural units (e.g., NH4+ in ammonium halides) and vacancy-induced disorder. They further suggested that

9464 J. Phys. Chem. B, Vol. 103, No. 44, 1999

Xue and Schrader

Figure 11. RI-VOPO4 phase transformations in wet air (5-10% water vapor/air) at 450 °C: (a) spectrum taken at RT; (b) after being heated to 450 °C; (c) after 2 h at 450 °C; (d) spectrum taken at RT after a thermal treatment for 4 h. Figure 10. δ-VOPO4 (with trace amounts of RII-VOPO4) phase transformations in wet air (5-10% water vapor/air) at 450 °C: (a) spectrum taken at RT; (b) after being heated to 450 °C; (c) after 1 h at 450 °C; (d) after 3 h at 450 °C; (e) spectrum taken at RT after a thermal treatment for 6 h.

massive vacancy-induced defects might totally destroy the translational symmetry and create a Raman scattering continuum. This argument is consistent with the absence of Raman bands observed during the transformation of VOHPO4‚0.5H2O to (VO)2P2O7.11 For the VPO catalyst transformations examined in this study, both types of disorder are likely involved. The bands near 1016-1020 and 993 cm-1 emerged after the total absence of Raman signals. Formation of “isolated” vanadium oxide species is likely. Such species for supported vanadia at low loadings have Raman bands in the region of 1016-1037 cm-1.28 Apparently the VPO materials experienced a structural disturbance due to the presence of water which likely involved the cleavage of V-O-P bonds; separation of the vanadium oxide species and phosphate species resulted. Similar to supported vanadia catalysts at higher loading,29 Raman bands due to V2O5 (e.g., 993, 703 cm-1, etc.) emerged and intensified, presumably as a result of the transformation of isolated vanadium species to form V2O5. Previous results from in situ high-resolution electron microscopy (HREM) imaging of (VO)2P2O7 under reaction conditions provide further insights into these transformations.13,30 Figure 12b shows the (010) plane of (VO)2P2O7 which exhibits a layered structure consisting of pairs of edge-sharing VO6 distorted octahedra equatorially linked together by cornersharing phosphate tetrahedra. Along the b-axis, the VO6 octahedra pairs are connected through interlayer bonding involving short and long vanadium-oxygen bonds (VdO‚‚‚ VdO) to form double columns. Phosphate tetrahedra in the two neighboring layers form pyrophosphate groups (with P-O-P bonds). Gai et al.13 monitored the behavior of (VO)2P2O7 samples treated in steam at ∼390 °C for 312 h using HREM:

extended defects along 〈201〉 directions were formed. A stepwise transformation13,30 involving gliding planes was proposed to explain this phenomenon: (1) in a reducing atmosphere (VO)2P2O7 loses basal oxygen which connects corner-sharing vanadium octahedra and phosphate tetrahedra (see Figure 12a); (2) oxygen vacancies are created and diffuse into the crystal (Figure 12b); (3) to reduce the strain generated by the misfit, the crystal lattice glides along 〈201〉 directions, and extended defects are formed (Figure 12c). The 〈201〉 defects were also formed for n-butane, N2, and H2 treatment, and the relative defect concentrations were related to the reducing power of the gases. Treatment in n-butane created more defects than did N2 or H2 treatment. An n-butane/air mixture generated the lowest defect concentration, evidently as the result of a lower number of oxygen vacancies. The defect concentration was higher in samples treated in steam than in those reacted in n-butane and N2 under similar conditions. After extended treatment in steam, HREM images showed regions covered by defects which might be depicted as in Figure 12d. We propose that additional V-O-P bonds in the defect region can be broken through further loss of oxygen atoms. The vanadium oxide species and phosphate groups can be separated, leaving chainlike structures of vanadium oxide. These vanadium oxide species then form a more stable structure by shifting along the a-axis (Figure 12e). The resulting lattice (Figure 12f) is converted to V2O5 (Figure 12g) after a structural rearrangement to make VdO bonds in every edge-sharing vanadium octahedra pairs “trans” to each other. It is therefore plausible that a (VO)2P2O7 crystal with high density of defects along 〈201〉 directions can be converted to V2O5 after a prolonged treatment in the presence of water vapor which removes V-O-P oxygen atoms. The Raman spectra for (VO)2P2O7 conversion to V2O5 are consistent with this model. The loss of Raman signal can be attributed to a large number of defects induced by the loss of

VPO Catalyst Transformations J. Phys. Chem. B, Vol. 103, No. 44, 1999 9465

Figure 12. Proposed pathway for (VO)2P2O7 transformation to V2O5: (a) top-basic structural unit of (VO)2P2O7, bottom-projection of basic structural unit (only oxygen atoms are shown); (b) (010) plane of (VO)2P2O7 with oxygen vacancies; (c) formation of a defect in 〈201〉 direction; (d) region covered by defects; (e) resulting vanadium oxide species after phosphorus oxides being removed; (f) vanadium oxides forming a close-packed structure; (g) (010) plane of V2O5.

9466 J. Phys. Chem. B, Vol. 103, No. 44, 1999 V-O-P oxygen atoms. The emergence of bands due to isolated vanadium oxide species (1016-1020 cm-1) may correspond to the stage when V-O-P bonds have been broken and vanadium groups and phosphate groups have become separated. The growth of V2O5 bands in Raman spectra might correlate with observation of the formation of a new phase by Gai et al.13,30 The building blocks for RI-, RII-, β-, δ- and γ-VOPO4 are also VO6 octahedra and PO4 tetrahedra.19 Structural differences arise from variations in connections (e.g., V-O-P bonds) and arrangement of these units. The band near 1016-1020 cm-1 (isolated vanadium oxide species) has been observed in conversions from the VOPO4 phases to V2O5 which suggests that these transformations may involve similar steps as for (VO)2P2O7 conversion. The preferential cleavage of V-O-P bonds during the reaction is supported by evidence obtained from experiments involving reduction-oxidation step changes for (VO)2P2O7 (Figure 2). Raman signals due to two types of linkages can be identified in the high wavenumber region: the pyrophosphate P-O-P stretching band at 934 cm-1; the V-O-P stretching bands of (VO)2P2O7 at 1182 and 1136 cm-1; and those due to V-O-P bonds in VOPO4 phases in the region of 1000-1100 cm-1. The P-O stretching bands in VOPO4 phosphate groups are very close to the 934 cm-1 band and were presumably obscured by this band. In oxidation steps, the vanadium pyrophosphate V-O-P linkages decreased while the vanadium orthophosphate V-O-P linkages were formed as indicated by changes in the intensity of their corresponding Raman bands. In reduction steps, the bands due to vanadium orthophosphate V-O-P stretching vibration were eliminated, indicating the breaking of V-O-P bonds as a result of reduction. In general, the pyrophosphate V-O-P linkage was also weakened during reaction as evidenced by the decrease in intensity of the Raman bands at 1182 and 1136 cm-1. The cleavage of V-O-P bonds also produces phosphorus species. Conversion of VPO catalysts to V2O5 should involve the removal of phosphorus species from the catalyst lattice. The migration of phosphorus compounds from the bulk to the catalyst surface in the presence of water vapor was recently reported by Richter et al.31 In their study, the surface P/V ratio of (VO)2P2O7 was monitored using XPS and ion scattering spectroscopy. For increasing treatment time in water vapor (20% H2O in N2 at 480 °C), the surface P/V ratio increased continuously from around 1.3 initially to about 1.5 after 10 hours of treatment. When the sample was subsequently left in a dry N2 stream, the surface P/V ratio decreased. Investigations using thermogravimetry further revealed weight losses for (VO)2P2O7 in wet N2 at 400 °C and 550 °C (not at 300 °C). Phosphoric acid was detected in the condensate from the reactor. Catalyst weight loss did not occur in dry N2 under similar conditions. In our studies the postreaction samples were characterized by IR spectroscopy to provide information about the phosphorus species in the catalysts. Figure 13a shows a portion of the IR spectrum for a postreaction sample whose Raman spectrum indicated a mixture of β-VOPO4 and V2O5. Figure 13 also provides a comparison of the IR spectra for pure β-VOPO4 (Figure 13b) and V2O5 (Figure 13c) in the same region. Features for both β-VOPO4 (1164, 1056, 1000, and 940 cm-1) and V2O5 (1020 cm-1 and a shoulder band around 820 cm-1) were apparent. A shoulder band around 1082 cm-1 in Figure 13a has been observed repeatedly in IR spectra for postreaction samples starting with different VPO materials. On the basis of IR spectra of vanadium phosphorus oxides and inorganic phosphorus compounds, the 1082 cm-1 band might be assigned to an ionic

Xue and Schrader

Figure 13. IR spectrum of (a) a (VO)2P2O7 sample after treatment in wet feed; (b) reference β-VOPO4; (c) reference V2O5.

phosphate (PO43-). This assignment is justified since (1) the PO43- has a IR active ν3 fundamental vibration frequency at 1082 cm-1;32 (2) the stretching vibration frequency of the V-O linkage is usually at lower wavenumbers; and (3) analysis of the IR spectra for numerous postreaction samples failed to find signals due to P-OH linkages. Water may be responsible for the migration of phosphoric compounds to the catalyst surface. Water molecules can diffuse into the lattice of VPO crystals at room temperature,19 and this process may occur more readily at higher temperatures. It may then react with isolated phosphorus oxide species and form acid phosphates which diffuse from the lattice matrix. IR spectra for postreaction samples did not reveal the presence of acid phosphate species. This may be because of the low concentration of acid phosphates and the ex situ nature of the characterization. A more definitive role for water cannot be suggested from the current study. For example, although a hydrolysis reaction with the V-O-P bond would suggest cleavage of these bonds and formation of phosphoric acid and V-OH, neither species could be directly observed. In contrast, the VPO phases cannot be readily converted to V2O5 in an oxidative atmosphere. The presence of gas-phase oxygen readily replenishes oxygen vacancies which hinders the cleavage of V-O-P bonds thus prevents the conversion of VPO phases to V2O5. Rather, all materials were converted to the most thermally stable β-VOPO4. The V-O-P oxygen appears to be the most reactive oxygen in the VPO catalyst lattice. Lashier and Schrader33 have found that O18 was preferentially incorporated into the lattice at V-O-P sites in the oxidation of (VO)2P2O7 to β-VOPO4. These V-O18-P oxygen atoms were active in the oxidation of n-butane which formed O18-rich maleic anhydride. In agreement with this prior observation, Raman spectra collected during the reduction-oxidation step change reactions in the current study revealed that the V-O-P bonds were among the first to be formed in the oxidation step, and they were also the most readily removed in the subsequent reduction step (Figure 2). Results by Gai et al.13,30 also clearly indicated that defects were formed as the result of the loss of V-O-P oxygen atoms in reducing atmosphere. The reactivity of the V-O-P oxygen might be attributed to low bond strength. On the basis of the calculation by Zio´lkowski et al.,34 the bond strength sum around the V-O-P oxygen is generally lower than that for VdO‚‚‚V and P-O-P oxygen atoms. The discovery made in this study suggests that the VPO catalyst deactivation may occur either as a result of catalyst oxidation to form V5+ phases or as a result of loss of

VPO Catalyst Transformations phosphorus, especially for those under frequent exposure to moisture in a reducing atmosphere. These observations may also provide the underlying reason for treatment of the VPO catalyst with a phosphorus compound in order to extend catalyst lifetime.35 Acknowledgment. The authors are grateful to K. Kourtakis and G. Coulston of DuPont Central Research and Development for synthesis of the catalyst used in these studies and for other helpful discussions. This work was conducted through the Ames Laboratory which is operated through the U.S. Department of Energy by Iowa State University under contract No. W-7405Eng-82. Support from the Office of Basic Energy Sciences, Chemical Sciences Division is also acknowledged. References and Notes (1) Arnold, E. W., III.; Sundaresan, S. Appl. Catal. 1988, 41, 225. (2) Lerou, J. J.; Mills, P. L. Du Pont Butane Oxidation Process. In Precision Process Technology; Weijnen, M. P. C., Drinkenburg, A. A. H., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1993; pp 175-195. (3) Contractor, R. M.; Horowitz, H. S.; Sisler, G. M.; Bordes, E. Catal. Today 1997, 37, 51. (4) Torardi, C. C.; Calabrese, J. C. Inorg. Chem. 1984, 23, 1308. (5) Johnson, J. W.; Johnston, D. C.; Jacobson, A. J.; Brody, J. F. J. Am. Chem. Soc. 1984, 106, 8123. (6) Leonowicz, M. E.; Johnson, J. W.; Brody, J. F.; Shannon, H. F., Jr.; Newsam, J. M. 1985, 56, 370. (7) Amoro´s, P.; Iba´n˜ez, R.; Martinez-Tamayo, E.; Beltra´n-Porter, A.; Beltra´n-Porter, D. Mater. Res. Bull. 1989, 24, 1347. (8) Amoro´s, P.; Iba´n˜ez, R.; Beltra´n, A.; Beltra´n, D.; Fuertes A.; GomezRomero, P.; Hernandez, E.; Rodriguez-Carvajal, J. Chem. Mater. 1991, 3, 407. (9) Beltra´n-Porter, D.; Beltra´n-Porter, A.; Amoro´s, P.; Iba´n˜ez, R.; Martinez, E.; LeBail, A.; Ferey, G.; Villeneuve, G. Eur. J. Solid State Inorg. Chem. 1991, 28, 131. (10) Bordes, E.; Johnson, J. W.; Raminosona, A.; Courtine, P. Mater. Sci. Monogr. 1985, 28B, 887.

J. Phys. Chem. B, Vol. 103, No. 44, 1999 9467 (11) Hutchings, G. J.; Desmatin-Chomel, A.; Oiler, R.; Volta, J. C. Nature 1994, 368, 4. (12) Bordes, E.; Courtine, P. J. Catal. 1979, 57, 236. (13) Gai, P. L.; Kourtakis, K.; Coulson, D. R.; Sonnichsen, G. C. J. Phys. Chem. B 1997, 101, 9916. (14) Bordes, E. Catal. Today 1987, 1, 449. (15) Moser, T. P.; Schrader, G. L. J. Catal. 1987, 104, 99. (16) Lashier, M. E.; Moser, T. P.; Schrader, G. L. In New DeVelopments in SelectiVe Oxidation; Centi, G., Trifiro`, F., Eds.; Elsevier Science Publishers: Amsterdam, 1990; pp 573-583. (17) Ben Abdelouahab, F.; Oiler, R.; Guilhaume, N.; Lefebvre, F.; Volta, J. C. J. Catal. 1992, 134, 151. (18) Bordes, E.; Courtine, P. J. Chem. Soc., Chem. Commun. 1985, 5, 294. (19) Ben Abdelouahab, F.; Volta J. C.; Olier, R. J. Catal. 1994, 148, 334. (20) Haggin, J. Chem. Eng. News, April 3, 1995, pp 20-23. (21) Soejarto, A. D.; Coulston, G. W.; Schrader, G. L. Can. J. Chem. Eng. 1996, 74, 594. (22) Centi, G.; Trifiro`, F.; Ebner, J. R.; Franchetti, V. M. Chem. ReV. 1988, 88, 55. (23) Horowitz, H. S.; Blackstone, C. M.; Sleight, A. W.; Teufer, G. Appl. Catal. 1988, 38, 193. (24) Guliants, V. V.; Benziger, J. B.; Sundaresan, S.; Wachs, I. E.; Jehng, J.-M.; Roberts, J. E. Catal. Today 1996, 28, 275. (25) Abello, L.; Husson, E.; Repelin, Y.; Lucazeau, G. Spectrochim. Acta 1983, 39A (7), 641. (26) Ali Omar, M. Elementary Solid State Physics: Principles and Applications; Addison-Wesley: Reading, MA, 1975. (27) Cheetz, B. E.; White, W. B. J. Am. Ceram. Soc. 1979, 62, 468. (28) Deo, G.; Wachs, I. E.; Haber, J. Crit. ReV. Surf. Chem. 1994, 4 (3/4), 141. (29) Went, G. T.; Oyama, S. T.; Bell, A. T. J. Phys. Chem. 1990, 94, 4240. (30) Gai, P. L.; Kourtakis, K. Science 1995, 267, 661. (31) Richter, F.; Papp, H.; Go¨tze, Th.; Wolf, G. U.; Kubias, B. Surf. Interface Anal. 1998, 26, 736. (32) Corbridge, D. E. C.; Lowe, E. J. J. Chem. Soc. 1954, 493. (33) Lashier, M. E.; Schrader, G. L. J. Catal. 1991, 128, 113. (34) Zio´lkowski, J.; Bordes, E.; Courtine, P. J. Catal. 1990, 122, 126. (35) Click, G. T.; Barone, B. J. U.S. Patent 4,515,899, 1985.